Richard Dronskowski

Hands-on quantum chemistry, in a seminar room of a nearby hotel. Despite spectacular summer weather, including a magnificent blue sky, the storm warning still exists, not taken down, so university and related institutes are locked... Thank you to private enterprise!

The participants of the 7th LOBSTER school, this year on Tenerife, at the IACTEC. Unfortunately, we brought the bad German weather with us. Apologies!

Der Übergang des Subjekts Jürgen Habermas in den Zustand radikaler Apophasis – jener Schwebe, in der das kommunikative Handeln nicht mehr als performativer Akt, sondern als stumme, unhintergehbare Kontingenz des leiblichen Substrats erscheint – vollzieht sich am 14. März 2026 in Starnberg als jene finale Dekonstruktion, die die Theorie selbst immer schon antizipierte: Die Unhintergehbarkeit des Arguments trifft auf die absolute Hintergehbarkeit des Argumentierenden, wodurch die idealisierende Unterstellung einer fortwährenden Diskursgemeinschaft sich nunmehr als kontrafaktische, posthume Projektion erweist. Was hier nicht endet, ist die Geltungssphäre des kommunikativen Vernunftpotentials; was hier lediglich suspendiert wird, ist dessen empirisches Trägerindividuum, das sich in die Anonymität der Lebenswelt zurückzieht, aus der es einst hervorging – ein performativer Akt der Selbstaufhebung, der die prekäre Rationalität des Öffentlichen nicht negiert, sondern in ihrer radikalen Endlichkeit allererst freilegt, sodass die nachmetaphysische Reflexion fortan ohne personalen Signifikanten, doch mit umso größerer normativer Dringlichkeit, weitergeführt werden muss. R.I.P Jürgen Habermas #Priebshow

I knew a reviewer (not me) who once wrote: "This paper should either be reduced by 50% or oxidized by 100%."

@TzinAt1521ce @bravo_abad Not that I know of, was not aware of that paper.

@bravo_abad @rdrons is anyone thinking of testing this with carbodiimides? Yours truly, handwritten letter circa 2015!

Predicting full optical tensors from crystal structure with an equivariant graph neural network Optical spectroscopy is one of the most direct ways to probe electronic structure in materials. The dielectric function—a frequency-dependent tensor describing how a material stores and dissipates electromagnetic energy—encodes everything from absorption edges to refractive indices, and its directional dependence governs applications in photodetectors, photovoltaics, and nonlinear optics. But computing this tensor from first principles is expensive: a single material requires DFT convergence followed by Brillouin-zone integration via the Kubo formula, taking ~30 minutes on hundreds of CPU cores. Scaling this to thousands of candidates is prohibitive. Ting-Wei Hsu and coauthors address this with TSENN: an equivariant graph neural network that maps crystal structures directly to full frequency-dependent dielectric tensors. The key insight is encoding the target not as six Cartesian spectra but via spherical harmonic decomposition—separating the isotropic trace (l=0) from the traceless anisotropic part (l=2). Because the architecture uses E(3)-equivariant tensor products, symmetries are respected by construction. The model never sees space group labels, yet predictions automatically satisfy the correct constraints: cubic materials yield identical diagonal components, hexagonal systems enforce xx=yy≠zz, and triclinic compounds retain all six independent elements. Trained on 1,432 semiconductors (0.3–3 eV band gaps), TSENN achieves a MAE of 0.127 on the imaginary dielectric tensor. Predicted spectra are accurate enough that real parts reconstructed via Kramers-Kronig relations closely match ab initio references. Strain tests confirm the model captures symmetry breaking quantitatively: uniaxial strain on a cubic material produces linearly growing anisotropy (R²=0.999), and shear distortion generates the expected off-diagonal component with sinusoidal dependence (R²=1.000). Inference takes ~1 second on a GPU versus ~30 minutes for DFT. The broader principle matters beyond optics: decomposing tensors into irreducible spherical representations with equivariant message passing extends naturally to piezoelectric or elastic tensors—a general recipe for learning symmetry-constrained responses from structure while bypassing expensive quantum-mechanical calculations. Paper: nature.com/articles/s4146…

A combined experimental/theoretical study identifying amorphous selenium as a highly effective ovonic threshold switch (OTS) material with ultra-low leakage, high on/off ratio, robust operation, and a trap-based charge-triggered mechanism: nature.com/articles/s4156…

@ChimiaDAO Thank you.

@rdrons This is really cool, congratulations!

The credit mostly goes to Hicham Bourakhouadar who carried out the synthesis, very precise work. This phase is a child, so to speak, of antiferromagnetic FeNCN and ferromagnetic Cr2(NCN)3, contributing to yet another enjoyable German-Japanese collaboration 🇩🇪🤝🇯🇵

Where is the "excess" electron in Ba2N = (Ba2+)2(N3-)e-? The singlet ground state has now been described using periodic "corresponding orbitals", showing where bonding and antibonding are taking place. pubs.aip.org/aip/jcp/articl…

We are staring into an abyss, and scientists are caught up in it, too; it is simply inconceivable. Jean Raspail saw the state of our world coming, and wrote about it artfully. The tragedy is that, although there is a German translation of the French original, there is no English one.

!!! Alena #Buyx unter #Plagiatsverdacht !!! 
Eine genaue Analyse ihrer #Doktorarbeit ergibt ein desaströses Bild. 
Nach Annette #Schavan und Mario #Voigt könnte sie als nächste ihren #Doktorgrad verlieren. Hier das ganze Gutachten: plagiatsgutachten.com/blog/wp-conten… Medizinische Doktorarbeiten aus der gleichen Forschergruppe können und dürfen sich ähneln, zumal im Material- und Methodenteil. Buyx hat aber wiederholt Literaturstudium vorgetäuscht, das andere schon vor ihr gemacht haben und dabei auch die Zitierfehler der anderen mit übernommen. Und nicht nur das, auch Fließtexte wurden abgeschrieben, sogar im Diskussionsteil. Das darf sie laut #KI und alter Methodenliteratur ganz eindeutig nicht. Wir fordern deshalb, dass sie auch ihre - wie wir annehmen: kumulative - #Habilitationsschrift der Öffentlichkeit jetzt nach 13 Jahren zugänglich machen muss. Denn erst diese hat sie als Ethikerin qualifiziert. #Corona Lesen Sie hier die ganze Story: plagiatsgutachten.com/blog/plagiatsv…

@simonmaechling All theoretical chemistry is really physics, and all theoretical chemists know it. - Feynman All many-body physics is really chemistry, and all theorists know it. - Anonymous

Chemistry is fun until someone tells you it's all physics.

Less than TEN days left to register for the free LOBSTER school on Tenerife, please check this website: schmeling.ac.rwth-aachen.de/lobster2026/in…

@0xlelouch_ Konrad Zuse Plankalkül, father of modern computers, Turing-complete Z3 John Backus FORTRAN, 1st widely used high-level programming language

Greatest software engineers of all time. 1. Alan Turing Father of computer science. Formalized computation itself. Without Turing Machines, there is no software engineering. 2. Dennis Ritchie Created C and co-created UNIX. Direct ancestor of Linux, Windows, macOS, Android, iOS. 3. Ken Thompson UNIX philosophy, file systems, processes, pipes. Software simplicity at its finest. 4. Linus Torvalds Linux kernel + Git. Literally runs the modern internet and developer workflows. 5. Donald Knuth Algorithms, complexity, correctness. The Art of Computer Programming shaped generations. 6. John von Neumann Von Neumann architecture - still how most computers execute software today. 7. Grace Hopper Invented compilers. Made software readable by humans instead of machines. 8. Claude Shannon Information theory. Compression, networking, error correction - all software relies on this. 9. Edsger Dijkstra Algorithms, structured programming, correctness proofs. Taught engineers how to think. 10. Brian Kernighan Software engineering discipline, tooling, and clarity. Made systems understandable. Modern-era system builders (honorable mentions) Jeff Dean - Distributed systems at planetary scale Martin Fowler - Software design patterns & refactoring Guido van Rossum - Python & developer productivity Any good ones I missed ??

The registration page for 7th LOBSTER chemical-bonding school (Tenerife, March 18-20, 2026) is now open! Please register here: schmeling.ac.rwth-aachen.de/lobster2026/in… See you on Tenerife! And please retweet! Deadline: February 6

... open!

The photos from ISSSC 2025 are here, big banquet: Ken, me, Jean-Marie, and Hiroshi eagerly awaiting the moment when we get to tap the sake barrel. It was great fun!

Honeycomb lattices, central to cathode materials and Kitaev physics, have a rare 3D analogue: the hyperhoneycomb lattice. Experiments and calculations show high pressure stabilises this structure. Reported in @J_A_C_S. @JST_ASPIRE with Bordeaux and Aachen. pubs.acs.org/doi/10.1021/ja…

What a wonderful Christmas surprise! 10 minutes ago Tsinghua University Press has informed me that the book has been published. Warm regards to China, all the best for 2026!